{"id":"721a4d8d-6284-4d9e-a280-fb9ea6a77a4b","arxiv_id":"2507.06371","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"THz pulses induce a millisecond-lived metastable magnetization in FePS3, with magnitude and lifetime growing near the Néel temperature through critical fluctuations.","lead":"Intense terahertz light pulses flip FePS3, an antiferromagnet, into a state with a net magnetization that lasts over 2.5 milliseconds, vastly longer than typical photoinduced changes. The effect is strongest near the material's magnetic transition temperature, suggesting that critical fluctuations can stabilize long-lived hidden states, a principle that may apply to other quantum materials.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2.5 ms magnetization lifetime is extracted from the ellipticity channel, not from the direct CD probe; a ms-scale nonmagnetic contribution or a saturating pre-time-zero ratio could mimic the claimed critical divergence.","rationale":"The reader's weakest assumption concerns Δη0(T) as a temperature-independent proxy for the magnetization amplitude. I agree that this is a real risk, but I think the more severe version of the same proxy problem applies to the lifetime: the direct CD measurement is not used for the ms-lifetime extraction, so the central statement of a 2.5 ms metastable magnetization is one step removed from the direct magnetization channel. The evidence for the existence of a CD response at 118 K and its absence under off-resonant driving is credible and is not being contested here. The concern is quantitative support for the divergence of the lifetime and the critical-slowing-down mechanism. The Fig. 3b power-law fits also lack quoted uncertainties, and the Methods note that the ratio-based extraction overestimates τdecay because the post-zero maximum is not reached; near TN this overestimation is temperature dependent because the rise is slower. A direct multi-τprobe CD measurement at 5 ms pump spacing would settle whether the ellipticity lifetime is actually the magnetization lifetime. This keeps the reader's CONDITIONAL verdict but sharpens the condition that must be met before the central claim is accepted.","tokens_in":21716,"tokens_out":17465,"duration_ms":219441,"concrete_test":"Measure the pre-time-zero ΔCD offset as a function of τprobe (1, 2, 3, 4 ms) with τpump = 5 ms at T = 118 K and at two temperatures on each side of TN, and fit the data with Eq. (4) exactly as was done for Δη in Fig. 3d. If the CD-based τdecay(T) agrees with the ellipticity values and follows the same power law, the magnetic lifetime and its critical divergence are directly confirmed. If the CD pre-time-zero offset is absent or gives a much shorter or non-diverging τdecay, the ellipticity channel contains a nonmagnetic ms component and the central claim fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The most load-bearing assumption is that the ms decay measured in the ellipticity channel is the decay of the magnetization itself. The direct magnetization probe, the THz-induced circular dichroism ΔCD, is only reported on sub-nanosecond time traces (Fig. 2g) and as a temperature-dependent amplitude ΔCD0(T) (Fig. 2h). The 2.5 ms lifetime is obtained by fitting the pre-time-zero ellipticity offset with the single-exponential pump-accumulation model, Eqs. (4)-(5), in Fig. 3c,d. This is an indirect inference: the paper itself states that Δη contains a non-thermal component beyond the static thermal response (Fig. 2f), so a slow lattice or thermal relaxation in the ellipticity channel could masquerade as a magnetic lifetime. The Methods also acknowledge that the ratio used for Fig. 3b overestimates τdecay because the post-zero maximum is not reached within the measured window; since the rise slows down near TN, this temperature-dependent overestimation can produce an apparent divergence even if the true magnetic lifetime does not diverge. The similarity between Δη0(T) and ΔCD0(T) supports the amplitude assignment only, not the ms decay.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports that intense broadband terahertz pulses resonantly driving low-energy magnon and phonon modes in the zigzag antiferromagnet FePS3 produce a long-lived state with a net out-of-plane magnetization near the Néel temperature (TN≈118 K). The evidence includes temperature-dependent polarization rotation, ellipticity, and circular dichroism transients, with a pre-time-zero ellipticity accumulation modeled as pump-pulse accumulation and fitted to yield a decay time of about 2.5 ms at 118 K. The authors attribute the state to nonlinear excitation of the 3.27 THz phonon mode, which modulates exchange couplings and, through a Ginzburg-Landau coupling g L M Q2, stabilizes finite M. First-principles DFT, Monte Carlo, and spin-dynamics simulations are used to support the microscopic mechanism, and critical exponents extracted from the temperature dependence are compared with the 3D Ising universality class.","tokens_in":21941,"tokens_out":9272,"duration_ms":100296,"significance":"If correct, the result would be a rare example of a THz-induced metastable magnetic state with millisecond lifetime, and the proposed mechanism—critical fluctuations of the dominant AFM order stabilizing a sub-dominant magnetization—is conceptually novel and could guide searches for hidden states near critical points. The paper is strong in its combination of direct time-resolved probes, a first-principles parameterization of exchange and spin-phonon couplings, and multiscale dynamical modeling; the model parameters are not fitted to the THz-induced signals, and the off-resonant control and direct CD measurements strengthen the amplitude assignment. However, the headline lifetime and its divergence are extracted from the ellipticity channel under an acknowledged overestimation, and the mapping between ellipticity and magnetization is not established on the millisecond timescale. These issues are load-bearing for the central claim and require additional evidence.","major_comments":[{"comment":"The 2.5 ms lifetime is extracted exclusively from the ellipticity channel via the pump-accumulation model of Eq. (4), while the direct magnetization probe (ΔCD) is reported only on sub-nanosecond traces (Fig. 2g,h). This is load-bearing because the manuscript states (Fig. 2f) that Δη contains a non-thermal contribution beyond the static thermal response; a slow lattice or thermal relaxation in the ellipticity channel could produce the same pre-time-zero accumulation without a magnetic origin. In addition, the Methods acknowledge that τdecay obtained from Eq. (5) is an overestimate because the post-zero maximum is not reached within the 30 ps window. Since the rise time itself slows near TN (Fig. 2i), the temperature-dependent overestimation can produce an apparent divergence in Fig. 3b even for a temperature-independent true lifetime. The authors should either measure ΔCD (or another direct magnetization probe) on millisecond time delays, or quantitatively rule out non-magnetic contributions to the slow ellipticity decay; without this, the central claim of a ms-lived metastable magnetization is not fully supported.","section":"Methods: Extraction of relaxation time; Fig. 3"},{"comment":"The derivation of M ∼ √χ_zz is not established in the main text. Minimizing Eq. (2) over M and Q2 for a fixed L gives M² = (g²L²/Ω² − a_M)/b_M, so M is proportional to sqrt(L² − L_c²), not to sqrt(⟨L²⟩) in general. Below TN, where L has a nonzero condensed value with L² ∼ |T−T_N|^{2β}, this would give an exponent β (or a shifted square-root onset), not γ/2; above TN the relation requires treating L as a fluctuating variable with ⟨L²⟩ ∼ χ_zz. The text should define the averaging and state explicitly why the condensed part can be neglected (or why the fluctuation part dominates in the measured temperature range). Since Eq. (3) and the comparison with the 3D Ising exponent in Fig. 2e rest on this relation, the theoretical support for the critical amplitude divergence is currently incomplete.","section":"Ginzburg-Landau theory, Eq. (2)-(3)"},{"comment":"The power-law fit of Δη0(T) to Eq. (6) assumes a temperature-independent proportionality between the measured ellipticity change and the induced magnetization M(T). The paper supports this by the similarity of Δη0(T) and ΔCD0(T), but both quantities are measured at the same short delay (~170 ps) and therefore cannot validate the long-lived component, and Fig. 2f shows that Δη contains an additional non-thermal contribution whose temperature dependence is not characterized. A temperature-dependent non-magnetic contribution to Δη near TN would change the fitted exponent and the inferred divergence. The authors should either isolate the magnetic component of Δη0(T) (e.g., through the CD channel with comparable statistics) or provide an explicit calibration of the ellipticity-to-magnetization conversion as a function of temperature.","section":"Results: 'we assume that Δη0(T) acts as a probe of the magnetization'; Fig. 2e"}],"minor_comments":[{"comment":"The fitted low-temperature decay exponent νz=0.72 (and 0.56 above) is described as 'close' to the 3D Ising value 1.27; with no error bars given for these values, this agreement claim should be softened or quantified.","section":"Methods: Fitting and extracting critical constants"},{"comment":"The vertical axis label should specify units of τdecay (ms) and indicate that the plotted values are overestimates, as stated in the text; currently the reader could mistake them for quantitative lifetimes.","section":"Fig. 3b"},{"comment":"The text alternates between Ω and Q for phonon labels (Ω1, Ωm, Ω3 vs. Q1, Q2); unify the notation to avoid confusion.","section":"Throughout"},{"comment":"Data availability is limited to 'on reasonable request'; for a study with strong computational and experimental components, a public data/code repository would improve reproducibility.","section":"Data Availability"},{"comment":"The paragraph introducing Eq. (8) uses 'critical constants' where 'critical exponents' is meant; correct the terminology.","section":"Methods: Fitting and extracting critical constants"}],"recommendation":"major_revision","confidential_remarks":"To the editor: This is a strong experimental-theoretical study, but the headline ms-lifetime claim depends on an indirect and acknowledged overestimate. I would not reject on current evidence, but I would ask for a direct ms-scale magnetization measurement or a quantitative decomposition of the ellipticity signal before publication. The novelty and potential impact are high; the requested experiments are within the authors' demonstrated capabilities."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here is my read. The core observation is real and new: intense THz driving near TN produces a magnetization in FePS3 that outlives the pump by milliseconds, and the effect strengthens as T approaches TN. If correct, that is a big step beyond the usual ps-ns photoinduced phases. The paper does several things right. The direct CD measurement at sub-ns timescales shows a transient net out-of-plane magnetization, the off-resonant control rules out a generic thermal/electronic artifact, and the Ginzburg-Landau mechanism—nonlinear Q2 phonon displacement modulating exchange, M proportional to sqrt(chi_zz), with critical slowing down inherited from the AFM order parameter—is backed by first-principles parameters that are validated against equilibrium neutron data and by Monte Carlo/spin dynamics. The exponent gamma/2 around 0.56 is consistent with 3D Ising. This is not a fitting exercise; the theory is parameterized before comparison.\n\nThe soft spots are real but not fatal. The 2.5 ms lifetime is extracted from the ellipticity channel, not from the direct CD probe. The paper itself says Delta-eta contains a non-thermal component beyond static thermal response, so a slow lattice or thermal relaxation in that channel could masquerade as a magnetic lifetime. The tau_pump = 5 ms, tau_probe-dependence study is a good control, but it still lives in the same optical channel. And the Methods note that the extracted tau_decay overestimates the true decay because the post-zero maximum is not reached in the window; since the rise slows near TN, that overestimation is temperature-dependent and can produce an apparent divergence even if the true lifetime does not diverge. Amplitude-wise, the similarity between Delta-eta0(T) and Delta-CD0(T) is reassuring, but it only validates the amplitude assignment, not the millisecond decay.\n\nThere are also internal inconsistencies: the Fig. 3b caption reports nu-z = 0.89/0.56 while Methods reports 0.72/0.56 for the same fits, and the decay-time exponents are quoted without uncertainties. These need fixing. The mapping from ellipticity to magnetization is explicitly assumed; a temperature-dependent nonmagnetic contribution would compromise the extracted exponent.\n\nBottom line: this paper deserves a serious referee. The central claim is important and the evidence is suggestive but not airtight. It would benefit from direct ms-scale CD measurements or another magnetization-sensitive probe (e.g., time-resolved XMCD) to confirm the long lifetime. I would send it to review with a request for clarification.","headline":"Genuinely new millisecond-scale photoinduced magnetization near TN in FePS3 with a plausibly parameterized mechanism, but the long lifetime rests on an indirect optical proxy and the reported critical exponents have internal inconsistencies.","tokens_in":22573,"tokens_out":2143,"would_cite":true,"duration_ms":23970,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A terahertz pulse creates a millisecond-lived magnetization in FePS3 near its critical temperature.","keywords":["FePS3","metastable magnetization","terahertz excitation","critical fluctuations","phonon-induced magnetism","van der Waals antiferromagnet","Ginzburg-Landau theory","critical slowing down"],"falsifier":"A decisive check would be time-resolved X-ray magnetic circular dichroism at the iron edge after the terahertz pump at 118 K: if no dichroism with a decay time near 2.5 ms appears, or if its temperature dependence does not diverge with exponent $\\gamma/2 \\approx 0.56$, then the ellipticity signal is not the paper's magnetization. A second check is time-resolved X-ray diffraction to verify that the 3.27 THz phonon displacement pattern is present and that the induced magnetization scales quadratically with terahertz field.","tokens_in":21474,"feed_emoji":"🧲","tokens_out":8958,"duration_ms":90882,"temperature":0.7,"pith_summary":"Terahertz light, not heat, can create a hidden magnetic state in the layered antiferromagnet FePS3 that survives long after the pulse is gone. The paper shows that an intense terahertz pulse resonantly drives a 3.27 THz phonon mode, and the phonon's displacement changes the distances between magnetic ions in a way that favors a net out-of-plane magnetization. That magnetization is metastable for more than 2.5 milliseconds, and both its amplitude and its relaxation time grow sharply as the temperature approaches the 118 K Néel point. The growth follows a power law with exponents consistent with three-dimensional Ising critical behavior, indicating that the antiferromagnet's own critical fluctuations amplify and stabilize the new state. If this is right, regions near phase transitions are places to look for long-lived light-induced phases.","feed_headline":"Terahertz pulses create a 2.5-ms magnetization in FePS3","feed_subtitle":"Critical fluctuations amplify this light-induced magnetic state and stretch its life to milliseconds.","key_machinery":"The load-bearing object is the 3.27 THz phonon mode $Q_2$ and its coupling $g L M Q_2$ to the antiferromagnetic order parameter $L$ and the magnetization $M$. This trilinear term makes the magnetic free energy develop two shallow degenerate minima at finite $\\pm M$ when the phonon is displaced; driving the phonon selects one minimum, and the sign of $M$ is fixed by $Q_2 L$. The same coupling makes $M$ relax adiabatically with $L$, so the metastable magnetization inherits the critical slowing down of the antiferromagnetic order parameter, $\\tau \\sim |T - T_N|^{-\\nu z}$, which is why the lifetime reaches milliseconds near $T_N$.","core_discovery":"The central claim is that nonlinear resonant driving of a specific phonon mode, not sample heating, produces a metastable magnetization in FePS3. The 3.27 THz phonon displacement modulates nearest-neighbor exchange couplings, and through the Ginzburg-Landau coupling term $g L M Q_2$ it makes a state with finite magnetization $M$ energetically favorable while the zigzag antiferromagnetic order $L$ remains dominant. The induced magnetization is detected as a transient circular dichroism and ellipticity change of an 800 nm probe, and its sign is locked to the product $Q_2 L$, so it is not reversed by an external magnetic field. Near the Néel temperature, the amplitude grows as $M(T) \\sim |T - T_N|^{-\\gamma/2}$ with $\\gamma/2 = 0.56 \\pm 0.05$, close to the three-dimensional Ising value, and the lifetime diverges because $M$ follows the critically slowed relaxation of $L$. First-principles spin-phonon couplings, Monte Carlo simulations, spin dynamics, and stochastic Ginzburg-Landau dynamics all support this picture.","pith_inferences":["Beyond the paper: if the critical-fluctuation amplifier is generic, then other van der Waals antiferromagnets with strong spin-phonon coupling, or strained versions of FePS3 with a shifted $T_N$, should show similar terahertz-induced metastable magnetization; this is a direct experimental test.","Beyond the paper: the predicted $M \\sim \\sqrt{\\chi_{zz}}$ scaling means that tuning $T_N$ by pressure, strain, or thickness should shift the divergence in a predictable way, allowing controlled adjustment of both amplitude and lifetime.","Beyond the paper: combining the millisecond lifetime with optical writing and reading suggests a route toward non-volatile spintronic memory elements, if the state can be erased by a second pulse or by heating through $T_N$."],"forward_implications":["The same resonantly driven phonon should be able to imprint a long-lived magnetization in other spin-phonon-coupled magnets, not only FePS3.","Materials tuned close to a magnetic critical point become natural targets for metastable light-induced phases, since critical fluctuations amplify small couplings and slow down relaxation.","The sign of the induced magnetization is fixed by the phonon displacement and the antiferromagnetic domain, so an external magnetic field cannot flip it.","The state is addressable on millisecond timescales, which makes it usable with slow probes such as transport, Hall effect, and X-ray magnetic circular dichroism."],"supporting_citations":[{"why":"Supplies the spin Hamiltonian and the magnon-phonon mode assignments on which the microscopic model is built.","marker":"[40]"},{"why":"Provides the exchange Hamiltonian and magnetic parameters used in the first-principles and Monte Carlo modeling.","marker":"[49]"},{"why":"Gives the temperature-dependent magnetic susceptibility whose critical divergence underlies the predicted $M \\sim \\sqrt{\\chi_{zz}}$ scaling.","marker":"[24]"},{"why":"Supplies the dynamic critical phenomena formalism used to predict the diverging relaxation time of the order parameter.","marker":"[54]"},{"why":"Provides the high-precision three-dimensional Ising critical exponents used to compare the measured power-law exponents.","marker":"[53]"},{"why":"Supplies the experimentally determined spin-exchange parameters that validate the first-principles magnetic model.","marker":"[35]"},{"why":"Identifies the magnon-phonon strong coupling in FePS3 used to assign the coherent modes in the terahertz spectra.","marker":"[38]"},{"why":"Provides the nonlinear ionic Raman scattering mechanism invoked to explain how the 3.27 THz phonon is nonlinearly driven.","marker":"[48]"},{"why":"Offers the infrared resonant Raman effect as the alternative nonlinear excitation pathway for the driving phonon.","marker":"[50]"}],"fun_headline_variants":["Light-induced magnetization in FePS3 lasts milliseconds","Terahertz writes a 2.5-ms magnetic memory in FePS3","Critical fluctuations help light create lasting magnetization","Phonon-driven metastable magnetism near Néel point","FePS3: Terahertz pulses give magnetism a 2.5-ms life"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the measured ellipticity change is a temperature-independent readout of the induced magnetization; if a non-magnetic contribution, such as a phonon-driven lattice distortion, adds to the ellipticity near $T_N$, the extracted critical exponent and the magnetic divergence claim would be compromised.","fun_headline_variants_meta":{"raw":{"variants":["Light-induced magnetization in FePS3 lasts milliseconds","Terahertz writes a 2.5-ms magnetic memory in FePS3","Critical fluctuations help light create lasting magnetization","Phonon-driven metastable magnetism near Néel point","FePS3: Terahertz pulses give magnetism a 2.5-ms life"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000643,"raw_usage":{"total_tokens":3000,"prompt_tokens":1033,"completion_tokens":1967,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":649,"completion_tokens_details":{"reasoning_tokens":1879}},"tokens_in":649,"tokens_out":1967,"duration_ms":14458,"temperature":1.0,"reasoning_tokens":1879,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:06:58.112915+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check would be time-resolved X-ray magnetic circular dichroism at the iron edge after the terahertz pump at 118 K: if no dichroism with a decay time near 2.5 ms appears, or if its temperature dependence does not diverge with exponent $\\gamma/2 \\approx 0.56$, then the ellipticity signal is not the paper's magnetization. A second check is time-resolved X-ray diffraction to verify that the 3.27 THz phonon displacement pattern is present and that the induced magnetization scales quadratically with terahertz field.","supporting_citations":[{"cited_title":"Chirality selective magnon-phonon hybridiza- tion and magnon-induced chiral phonons in a layered zigzag antiferromagnet.Nature Communications, 14(1), June 2023","cited_arxiv_id":null,"evidence_quote":"Provides the exchange Hamiltonian and magnetic parameters used in the first-principles and Monte Carlo modeling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the temperature-dependent magnetic susceptibility whose critical divergence underlies the predicted $M \\sim \\sqrt{\\chi_{zz}}$ scaling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the dynamic critical phenomena formalism used to predict the diverging relaxation time of the order parameter."},{"cited_title":"Pushing the limits of Monte Carlo simulations for the three-dimensional Ising model.Physical Review E, 97(4):43301, 4 2018","cited_arxiv_id":null,"evidence_quote":"Provides the high-precision three-dimensional Ising critical exponents used to compare the measured power-law exponents."},{"cited_title":"The magnon dynamics and spin exchange parameters of FePS3.Journal of Physics Condensed Matter, 24(41):8, 2012","cited_arxiv_id":null,"evidence_quote":"Supplies the experimentally determined spin-exchange parameters that validate the first-principles magnetic model."},{"cited_title":"Prosnikov, David Sedmidubský, Zdenek Sofer, Peter C.M","cited_arxiv_id":null,"evidence_quote":"Identifies the magnon-phonon strong coupling in FePS3 used to assign the coherent modes in the terahertz spectra."},{"cited_title":"Sum-frequency ionic Raman scattering.Physical Review B, 97(17):174302, 5 2018","cited_arxiv_id":null,"evidence_quote":"Provides the nonlinear ionic Raman scattering mechanism invoked to explain how the 3.27 THz phonon is nonlinearly driven."},{"cited_title":"Benedek, and Jeffrey Moses","cited_arxiv_id":null,"evidence_quote":"Offers the infrared resonant Raman effect as the alternative nonlinear excitation pathway for the driving phonon."}],"review_version":1}